Optimal Allocation Of Distributed Generation In Distribution Network Planning

The distribution network planning is a nonlinear, multi-objective complex systemmixed optimization problem. Locating and sizing of distributed generation(DG) accessto distribution network has become an important topic in distribution networkplanning. Distribution networks designed usually to closed-loop structure with multipleloop could run under different radial topology structure by different combinations ofsection switches and tie switches. The distributed generation would change thedistribution network flow and voltage distribution, network loss and node voltage arenot only related to the location and capacity of DG, but also are closely related to thestructure of the distribution network. Therefore, this paper takes access address andinstalled capacity of distributed generation and distribution network operating structureinto account to complete collaborative optimization planning.Firstly, this paper studies the models and treatment methods of different distributedgenerations in distribution network flow calculation. And proposes to get the initialvalue of Newton Raphson power flow calculation using Forward-backward sweepmethod after one iteration to ensure flow calculation converge reliably and fastly.Collaborative optimization model of distribution network and structure and addresscapacity DG is established, using Prüfer coded improved genetic algorithm to completeoptimization. According to the network characteristics of the tree structure in runtime, using Prüfer number to encode theoretically feasible distribution operation topologystructure and the integer to encode the access nodes and the installed capacity ofdistributed generation makes the distribution network running structure and the choiceof DG address-capacity combine into the evolution of the chromosomal gene selectionproblem. According to the characteristics of the Prüfer-coded, makes some restrictionsand improvements in crossover and mutation operation to solve the problem by binaryencoding that it is easy to produce illegal solution and repair difficultly in order toimprove computational efficiency and convergence rate. Finally, an example is given tovalidate the feasibility and advantage of this algorithm.In order to solve the output power instability problems of wind power andphotovoltaic power generation affected by natural conditions. Collaborativeoptimization model of the variable output power of wind power and photovoltaic poweris established. Use Weibull distribution to describe the random distribution of the wind speed, simplify the output power-wind speed characteristics of the wind generator toobtain the probability distributions of wind generator output power. For photovoltaicgenerator, establish the single light intensity as the independent variable power outputmodel influenced by sunny, cloudy, rain-snow three kinds of weather conditions andfour seasons for one cycle. To further improve the performance of the algorithm andsolve the lack of solution space smaller in the integer coding capacity of DG, thispaper proposes arithmetic crossover genetic algorithm improved by optimizationprinciple of particle swarm. Use improved algorithm to solve the problem ofcollaborative hybrid optimization between the distribution network structure andlocating and sizing of constant, variable out-power distributed generation. Theimproved algorithm still encode distribution network structure by Prüfer-coded modeland encode capacity of distributed generation by real-coded model. Then to analysis andverify the feasibility of this evolutionary algorithm by examples.